INTERNATIONALJOURNAL OF SYSTEMATICBACTERIOLOGY, July 1994, p. 527-533 Vol. 44, No. 3 0020-7713/94/$04.00+0 Copyright 0 1994, International Union of Microbiological Societies Evolutionary Relationships among Eubacterial Groups as Inferred from GroEL (Chaperonin) Sequence Comparisons ALEJANDRO M. VIALE,* ADRIAN K. ARAKAKI, FERNANDO C. SONCINI, AND MUL G. FERREYRA Departmento de Microbiologia, Facultad de Ciencias Bioquimicas y Famackuticas, Universidad Nacionul de Rosario, 2000 Rosario, Argentina The essential GroEL proteins represent a subset of molecular chaperones ubiquitously distributed among species of the eubacterial lineage, as well as in eukaryote organelles. We employed these highly conserved proteins to infer eubacterial phylogenies. GroEL from the species analyzed clustered in distinct groups in evolutionary trees drawn by either the distance or the parsimony method, which were in general agreement with those found by 16s rRNA comparisons (i.e., proteobacteria, chlamydiae, bacteroids, spirochetes, firmicutes [gram-positive bacteria], and cyanobacteria-chloroplasts).Moreover, the analysis indicated specific relation- ships between some of the aforementioned groups which appeared not to be clearly defined or controversial in rRNA-based phylogenetic studies. For instance, a monophyletic origin for the low-G+C and high-G+C subgroups among the firmicutes, as well as their specific relationship to the cyanobacteria-chloroplasts, was inferred. The general observations suggest that GroEL proteins provide valuable evolutionary tools for defining evolutionary relationships among the eubacterial lineage of life. The study of macromolecules emphasizing the historical MATERIALS AND METHODS information contained in their sequences has resulted in profound changes in our conception of the evolution of life on GroEL as an evolutionary chronometer. Widely different our planet, with its attendant consequences for the classifica- base compositions in the different lineages under study have tion of living organisms (27, 30, 43, 53). In particular, compar- been reported to constitute potential sources of inconsistencies isons of the 16s rRNA sequences from a large number of when nucleotide sequences (including those of rRNA) are species have been pivotal in providing evidence of three compared for phylogenetic studies (15, 30, 44). Since this primary lines of descent, two of them leading to the prokary- substitutional bias is minimized in highly conserved proteins otic lineages (Eu)Bacteria and Archaea (27,28,30,53). These (15, 20), inferences based on comparisons of their amino acid studies also indicate that the eubacterial lineage has evolved sequences have been proposed to be more reliable than those into (at least) 10 distinct divisions, although the specific based on the corresponding nucleotide sequences (15). relationships among (and sometimes within) them have yet to Analysis of the GroEL proteins of the organisms listed in be convincingly determined (28, 30, 31, 43, 53, 56). Table 1 indicates that the tendencies seen in nonconserved proteins (i.e., correlation of low G+C base content with Given the limitations inherent in the assumptions on which increases in Ile, Lys, Phe, and Tyr on the one hand and high current phylogenetic methods are based, phylogenies based on G+C content with Ala, Arg, and Gly on the other [20]) are a single macromolecule may not necessarily reflect the true minimized in these highly conserved proteins (49) and that phylogeny of the lineages in which it occurs (5,9, 15,30,43,44, most of the amino acid changes result in conservative substi- 56). Therefore, it is becoming increasingly evident that reso- tutions (10, 49). Moreover, the size (ca. 550 amino acid lution of the evolutionary relationships between organisms residues), as well as the highly conserved function, of these (especially prokaryotes) undoubtedly requires comparative proteins (10, 13) appears to include most of the desirable analysis of data from different macromolecules showing useful features of a molecular chronometer (30, 43, 53). Therefore, features as molecular chronometers (5, 30, 43). we used comparisons of GroEL amino acid sequences rather The GroEL, or Hsp60 (the common major antigen in than the corresponding nucleotide sequences for inferences of numerous eubacterial genera [7]), chaperonins constitute a eubacterial phylogenies. family of highly conserved housekeeping proteins. These pro- Data sources and data base searches. The organisms from teins are ubiquitously distributed among eubacteria and eu- which groEL genes have been characterized, their affiliations karyotic organelles and possess functions essential for the according to 16s rRNA analysis, and the sources of informa- survival of cells in physiological, as well as stressful, situations tion are provided in Table l. DNA and protein data base (7, 10, 13). The similarities between evolutionary trees drawn searches were performed at the National Center for Biotech- from a limited set of these molecules and those of 16s rRNA nology Information by using the BLAST network service (2). have been noted previously (7, 10). We extended this phylo- Data analysis. Alignments of the 58 GroEL protein se- genetic analysis by using an expanded GroEL data base and quences indicated in Table 1were done as described previously found that these proteins represent valuable molecular chro- (lo), and final adjustments were decided after visual inspec- tion. To calculate evolutionary distances, 525 aligned positions nometers. were employed after removal of ambiguous alignments that include in all sequences a stretch of nine amino acids equiva- lent to Escherichia coli GroEL positions 427 to 435), the C-terminal portion (starting at the position equivalent to E. coli GroEL position 531), and transit peptides from eukaryotic * Corresponding author. Phone: (54-41) 821701. Fax: 54-41-300309 sequences. Evolutionary distances were computed by using the or 54-41-240010. Electronic mail address: [email protected]. amino acid conversion table (PAM 001) compiled by Dayhoff 527 528 VIALE ET AL. INT.J. SYST.BACTERIOL. TABLE 1. Organisms, corresponding affiliations, and sources of TABLE 1. Continued Hsp60 sequences used in this work Organism and affiliation" GenBank Reference accession no. Organism and affiliation" GenBank Reference accession no. a-Proteobacteria Cyanobacteria Bartonella bacillifomzis M98257 UP Synechococcus sp. strain PCC 7942 M5875 1 10 Brucella abortus LO9273 14 Synechocystis sp. strain PCC 6803" D12677 10 Agrobacterium tumefaciens X68263 39 Synechocystis sp. strain PCC 6803 M57517 22 Rhizobium leguminosarum L20775 UD Rhizobium meliloti A" M94192 35 Chloroplasts Rhizobium meliloti C M94190 35 Cyanidium caldarium X62578 24 Rhizobium meliloti c" M94191 35 Triticum aestivum ad X0785 1 16 Bradyrhizobium japonicum 2' 222604 11 Ricinus communis (Y X07852 16 Bradyrhizobium japonicum 3" 222603 11 Brassica napus 01 M35599 10 Zymomonas mobilis L11654 UD Brassica napus pd M35600 10 Ehrlichia chaffeensis L10917 46 Arabidopsis thaliana p JT0901" 55 Rickettsia tsutsugamushi M31887 45 " As indicated by 16s rRNA analysis (28, 31). P-Proteobacteria UD, unpublished data. Neisseria gonon-hoeae 223008 UD groEL is present in a groESL operon (this applies only in cases in which more Neisseria flavescens 222955 UD than one groEL gene has been reported in a particular species). 'a/p indicates the type of polypeptide that composes the plant chloroplast Neisseria meningitidis 222956 UD GroEL chaperonin (10). PIR accession number. y-Proteobacteria Coxiella burnetii M20482 10 Pseudomonas aeruginosa M63957 42 Haemophilus ducreyi M91030 32 et al. (8). For construction of phylogenetic trees, we employed Salmonella typhi U01039 UD the neighbor-joining distance method (36), which has been Escherichia coli X07850 16 shown in model studies to be relatively consistent compared to Symbiont P of Acyrthosiphon pisum X61150 10 other methods, even in the presence of unequal rates of Yersinia enterocolitica X68526 UD evolution (36). For comparisons, phylogenetic trees were also Chromatiurn vinosum M99443 10 inferred by using the PROTPARS maximum-parsimony Legionella pneumophila M31918 18 Legionella micdadei X57520 10 method (9). Confidence limits for the inferences obtained were Symbiont of Amoeba proteus M86549 1 placed by using the "bootstrap" procedure (9). The programs PROTDIST, NEIGHBOR, PROTPARS, WE-Proteobacteria SEQBOOT, and CONSENSE, present in the PHYLIP pack- Helicobacter pylon' X73840 23 age (9), version 3.5 (kindly provided by J. Felsenstein, Univer- sity of Washington, Seattle), were employed in this work. Bacteroids Porphyromonas gingivalis D17398 UD GroEL protein alignments and evolutionary distances were provided for the reviewing process and are available from us Chlamydiae on request. Chlamydia trachomatis M31739 10 Chlamydia pneumoniae M69217 21 Chlamydia psittaci X5 1404 10 RESULTS AND DISCUSSION Spirochetes The evolutionary relationships between the eubacterial spe- Leptospira interrogans L14682 3 cies listed in Table 1, as inferred from their GroEL proteins, Borrelia burgdoiferi W4059 40 are shown in Fig. 1. Our analyses by the distance (Fig. 1) and Treponema pallidum X54111 17 parsimony (data not shown) methods indicate the presence of two defined clusters, one of which includes proteobacteria, Firmicutes (gram-positive bacteria), chlamydiae, spirochetes, and the eubacterial species Porphy- low G+C content Clostridiumperjhngens X62914 10 romunas gingivalis and Helicobacter pyluri and appears